Ethical Behavior Analysis in the Age of Artificial Intelligence (AI): The Importance of Understanding Model Building while Formal AI Literacy Curricula are Developed.

Ethical Behavior Analysis in the Age of Artificial Intelligence (AI): The Importance of Understanding Model Building while Formal AI Literacy Curricula are Developed - PMC Skip to main content An official website of the United States government Here's how you know Here's how you know Search Log in Primary site navigation Search Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice Perspect Behav Sci . 2025 Jun 5;48(3):621–631. doi: 10.1007/s40614-025-00459-z Ethical Behavior Analysis in the Age of Artificial Intelligence (AI): The Importance of Understanding Model Building while Formal AI Literacy Curricula are Developed David J Cox 1, 2, ✉ Author information Article notes Copyright and License information 1 Institute for Applied Behavioral Science, Endicott College, Beverly, MA USA 2 Mosaic Pediatric Therapy, Charlotte, NC USA ✉ Corresponding author. Accepted 2025 May 2; Collection date 2025 Sep. © Association for Behavior Analysis International 2025. Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. PMC Copyright notice PMCID: PMC12411336  PMID: 40919363 Abstract Ethics is fundamentally concerned with claims of “right,” “wrong,” “good,” “bad,” and how we might know those claims are accurate. Artificial intelligence (AI) is a term that represents a suite of tools where nonbiological systems process data and information to generate an output considered “intelligent.” As with any suite of technologies, choosing the right tool requires the tool user to critically evaluate which tool is best for the task (if one exists). Harnessing the power of AI systems to maximize benefit and minimize harm requires basic AI literacy. AI literacy requires a basic understanding of how mathematical models function. All future behavior analysts will need to be AI literate. This will require changes to education and training programs to ensure students have a basic understanding of model building, especially as we wait for the scholarship and research to unfold that outlines AI literacy skills specific to behavior analysts. Keywords: Artificial intelligence, Behavior analysis, AI literacy, AI ethics, Quantitative models Artificial intelligence (AI) permeates nearly every modern human industry. Though much AI runs “in the background” of technological platforms (Haenlein & Kaplan, 2019 ), the Fall 2022 release of the chatbot interface to ChatGPT launched a revolutionary shift in modern society where anyone, anywhere, could use AI to influence many of the daily tasks they engage in. The success of ChatGPT is estimated to have generated $200,000,000,000 of investment in AI-related products (Goldman Sachs, 2023 ). Perhaps it is just the algorithms that feed me content in the digital space, but the AI industry seems to be expanding exponentially, with ads popping up for products I can buy that use a large language model (LLM) to do nearly anything I need. As someone whose career sits at the intersection of behavior analysis and AI, I find this both exciting and terrifying. On the one hand, it is astounding to watch sporting events, walk through the airport, or visit a local restaurant and see one’s professionally affiliated field advertised as making the experience or product better and more impressive than ever (and why it deserves your money instead of competing products that do not have AI). The terrifying bit comes from the ease with which people can now wield AI tools that emphatically claim to transform your life without the tool creator being transparent about the AI system’s accuracy, the model’s training relevance to the task at hand, or the AI system’s appropriateness to your current situation. In this time of rapid development and change, the question posed in this article is: How might behavior analysts maximize the benefits of using AI systems while minimizing or mitigating potential harms? 1 Or, asked differently, what skills might behavior analysts need to use AI ethically? To avoid setting unmet expectations, I note that I do not provide a prescriptive answer to these questions (e.g., AI should be used in these cases and not in these). This is done for two reasons. First, it is impossible to account for every AI tool that exists or will exist in the future. Second, broadly aligned with professional behavior, it seems more adaptive to have a foundational skillset that can be generalized to situations we encounter rather than memorizing a specific set of responses (e.g., Baldwin & Ford, 1988 ; Bransford et al., 2000 ). The quick answer to outlining the skills behavior analysts need to use AI ethically is to reference a burgeoning field termed AI literacy . Here, authors are actively defining exactly what makes someone AI literate (e.g., Lintner, 2024 ; Ng et al., 2021 ), the determinants of AI literacy (Celik, 2023 ), and how AI literacy skills can be made specific to individual industries (Domínguez Figaredo & Stoyanovich, 2023 ; e.g., education, Tan & Tang, 2025 ). The assumption is that people sufficiently educated about AI tools and how to use them will do so aligned with what they perceive as greater benefits and lesser harms (or can be held accountable for failure to do so). However, past research suggests that contingencies to support ethical AI use (and prevent misuse) will be needed in addition to simply AI education (e.g., Cox, 2021 ; Fauzihardani et al., 2019 ; Kish-Gephart et al., 2010 ; Reynolds et al., 2010 ). One challenge with bringing AI literacy to behavior analysis is that the exact skills needed to be AI literate are unknown. To be clear, many AI literacy checklists exist (see Lintner, 2024 , for a review). However, many of their psychometric properties are still unknown, and none seem likely to generalize easily to behavior analysis without work and research (Lintner, 2024 ). Thus, the growing AI literacy literature, extended upon by research and scholarship specific to behavior analysis, will likely help develop the skills and curriculum unique to AI literacy for behavior analysts over the long run. The challenge is that AI literacy in behavior analysis cannot wait: many researchers and practitioners need to improve their AI literacy today. AI products are flooding the research and practice landscape, with marketers and salespeople offering products that promise to reduce the effort of behavior-analytic work in exchange for a few dollars (e.g., Google “AI Research Platforms,” “AI-generated ABA session note summarizer,” “AI-generated ABA treatment plans”). This may be ethically 2 problematic because humans have long been known to trade off reduced accuracy for reduced effort (e.g., Johnson & Payne, 1985 ; Payne et al., 1988 ). And, ethical missteps are likely to increase as one’s ability to vet an AI system decreases. So, what can practicing behavior analysts, behavior analytic faculty, and students of behavior analysis do today to jumpstart their AI literacy skills? In what remains, I first outline the fundamental skill behavior analysts need to become AI literate, use AI ethically, and that will persist across rapid AI advancements. The logic runs as follows, with the lede thus far buried and referenced last. Moving forward, being an ethically effective behavior analyst means being AI literate. Being AI literate means you can critically evaluate which AI tools are appropriate to a task and the AI tool output at a level commensurate with your job role(s). Critically evaluating the appropriateness of an AI tool and its output requires a basic understanding of how it works. A generalizable and basic understanding of how AI works requires a basic understanding of how mathematical models function. Thus, ethically using AI requires a basic understanding of how mathematical models function as commensurate with one’s job role(s). Understanding Mathematical Model Building: The Core Skill Needed to Ethically Use AI AI is mathematics. There is no way to get around that fact. To understand AI at a basic level, you need to understand how models function and are used to build AI products. This does not mean that every behavior analyst needs to be capable of contributing to the Society for the Quantitative Analysis of Behavior ’s annual conference program (though how neat that would be!). Rather, it is about understanding the basics well enough that you can competently and critically evaluate the suite of AI tools available to you for use. We can use automobiles as an analogy to make this point. Most of us do not need to know the underlying technical mechanics of a car engine to use it safely in our everyday lives. We just need to know which vehicles we can afford, why we are purchasing or using a vehicle, what fuel to put into the gas tank so it runs as it ought to, when and how often to get the oil changed, how to operate it safely, how to behave on the road when others are using theirs, how to read the dashboard to safely operate it and get it fixed when something is wrong, and—maybe most important—whether we need to use a car at all (e.g., when doing your dishes). Understanding this information makes it easier for car users to choose the best car for their use case, operate it safely for themselves and those around them, and know when the car is not performing as it should or is not needed. We can translate the previous paragraph relative to AI systems. Most of us do not need to know the underlying technical mechanics of how every AI system operates to use them safely in our everyday lives. We just need to know which AI tools we can afford to buy, why we are purchasing or using an AI system, what data was went into the AI system and is therefore needed as input for it to run as it ought to, when and how often the AI system is maintained and updated, how to use it safely when only I am affected, how to use it safely when others’ lives are affected, how to read the AI system output to safely operate it and know when something is wrong, and—maybe most important—whether we need to use the AI system at all (e.g., when doing your dishes). The above analogy fails to capture the current AI landscape in an important way that is worth making explicit. Incredibly robust safety and standardization regulations surround automobile manufacturing. For example, very few worry about their car exploding when we start the ignition; the car falling apart when we reach 60 MPH; locating obscure fuel sources unique to our vehicle; or the pedals and steering mechanism being unique to different cars and requiring extensive training. In contrast, no widely accepted safety or standardization industry exists to regulate AI systems. Anyone can create, market, and try to sell anything, leading to potentially harmful consequences if misused (see Weapons of Math Destruction for how mathematical modeling can cause significant harm when used inappropriately; O’Neil, 2016 ). Without oversight to protect AI system users before AI systems are available on the market, each of us has to be independently capable of evaluating the safety, proper use cases, and quality of AI systems marketers are marketing and salespeople are trying to sell. So, what might this entail? To begin, a handful of skills are often discussed within the AI literacy literature 3 (Table 1 ; e.g., Celik, 2023 ; Domínguez Figaredo & Stoyanovich, 2023 ; Lintner, 2024 ; Ng et al., 2021 ; Tan & Tang, 2025 ; Walter, 2024 ). All these skills, in one way or another, rely upon or are enhanced by understanding how mathematical models function at a basic level. A parallel argument familiar to behavior analysts might make this point more salient. If you understand fundamentally how operant and respondent conditioning work, it is much easier to ethically assess behavioral function, design behavior change programs, and engage in conversation about implementing intervention programs compared to if you only know a topographical bag of behavior modification tricks. AI literacy curriculum will likely be developed for behavior analysts in the long run that specifies and provides resources for the many specific topographies associated with the characteristics of AI literacy in Table 1 . In the meantime, understanding how mathematical models function makes each characteristic of AI literacy obvious, including what it is and the related questions to ask when building or buying AI systems. Table 1. Published Characteristics of AI Literacy Characteristic Description Technical Understanding Basic knowledge of how AI systems work, including concepts like machine learning, neural networks, and the difference between narrow and general AI Critical Evaluation The ability to assess what an AI system is and is not good at, recognize potential biases, and question AI-generated outputs rather than accepting them uncritically Awareness of Applications Familiarity with how AI systems are or can be used across and within different domains (health care, finance, education) Data Literacy Understanding how data influences AI systems, including the concepts of data quality, data bias, data privacy, and how training data shapes AI system outputs Ethical Awareness Recognition of ethical considerations surrounding AI, including privacy, potential for discrimination, impact of automation, and appropriate use cases Societal Impact Awareness Understanding how AI affects social structures, employment, governance, and other broad societal factors Risk Assessment Ability to identify potential harms or unintended consequences of AI implementation in various contexts Practical Interaction Skills Ability to effectively communicate with and use AI systems as tools, including constructing appropriate prompts and interpreting responses Open in a new tab We can spell this out for a few examples. Developing a technical understanding of different AI systems (e.g., machine learning, neural networks, narrow and general AI, LLMs) is easier when you know they are different structures for building mathematical models. As another example, critically evaluating an AI system output (and what someone is trying to sell you) is much easier when you understand how the data used to build a model constrains the set of possible outputs and how to interpret model performance scores (e.g., accuracy, F1 scores, mean absolute error; Cox & Vladescu, 2023 ). Related to this, as described elsewhere (Jennings & Cox, 2024 ), no model or AI system is free from bias. Understanding how mathematical models function makes it easier to identify sources of bias in AI systems via data selection, feature engineering, or algorithm design. Practical Implications for Education and Training The argument being made is not that everyone should be a mathematical wizard who thinks in equations. Rather, the specific education and training-related details will depend on the level of understanding someone requires for their behavior analytic role(s). Table 2 provides a first pass at what AI literacy levels might look like, ranging from foundational awareness (e.g., RBTs, entry-level BCBAs) to technical fluency (e.g., academic researcher using AI in their work) or strategic leadership and governance (e.g., CEOs, COOs, academic department chairs). Note that each level focuses on a different level of detail, but that all rest on understanding how mathematical models function. Table 2. Types of AI Literacy Levels and Who They Might Apply to in Behavior Analysis Level Description Skills Context Roles Foundational Awareness Basic understanding of what AI is, common tools, and the ethical implications of their use • Recognize AI-generated content • Understand limitations and bias in AI tools • Aware of data privacy issues • Daily clinical documentation • Use of AI embedded within practice management software • Data visualization using AI-enhanced tools • RBTs • Entry-level BCBAs • Direct service providers • Master’s students Applied Proficiency Can interact with AI tools responsibly, use them to streamline documentation, generate materials, conduct basic analyses, or generate instructional content • Prompt engineering for clinical utility • Review and refine AI outputs for accuracy and alignment with evidence-based practices • Using AI to assist with treatment plan drafts, graphing, and admin work • Treatment planning • Supervision and documentation support • Curriculum design • Graduate coursework and thesis support • Mid-career BCBAs • Clinical Supervisors • Curriculum developers • Educator’s teaching master’s level courses • Practitioners engaged in quality assurance projects Analytical & Ethical Integration Ethical and analytical oversight of AI use in behavior analysis. Ensures alignment with professional standards (e.g., BACB, 2020 ) and integrates AI with critical reasoning and clinical-ethical decision-making models • Auditing AI decisions or summaries (e.g., red-teams) • Applying and teaching clinical-ethical decision-making related to AI use • Monitoring bias, fairness, and client safety related to AI outputs • Clinical program oversight • Multidisciplinary collaboration • Organizational policy development • Ethics education in graduate programs • Clinical Directors • Quality Assurance Professionals • Ethics Committees • Vendor due diligence • Educators teaching research methods or ethics Technical Fluency Deep understanding of how AI systems work, including model architectures, data pipelines, and algorithmic bias. Can influence design and evaluation of tools • Review, interpret, and evaluate algorithm performance and bias • Collaborate with developers to ensure behavioral relevance • Implementing internal policies on AI integration • Product development • Research collaboration with data scientists • Experimental analysis of behavior influenced by AI models • Tool development in industry or academic labs • BCBA-Ds with a focus on AI application • Data analysts/scientists with ABA organizations • Research scientist • Faculty teaching advanced quantitative or AI methods Strategic Leadership & Governance Leadership in guiding ethical AI adoption. Shapes policies, sets standards, and contributes to national/international discourse on AI governance • Establishing institutional and organizational AI use guidelines • Balancing innovation with client safety • Shaping the profession’s role in AI regulation • Shaping graduate curriculum and accreditation standards for AI literacy • Executive leadership of ABA orgs and professional associations • Policy developers • Higher education administration • AI governance panels and policy think tanks • CEOs/COOs of ABA companies • University program chairs and deans • Professional association leaders • Research center directors Open in a new tab So, where might someone start? If faculty teaching primarily in a master-level course sequence, I would start by examining each course our program offers and create a module or spend one week on AI tools specific to that course topic (see Cox & Jennings ( 2023 ) for a host of additional examples). For example, an assessment course could review how AI can aid single-subject data analysis (e.g., Turgeon & Lanovaz, 2020 ) or how an LLM can generate behavior intervention plans based on functional assessment results that the students could then critique. As a second example, a theory course could involve a module where students interact with LLMs around behaviorism to critique its accuracy or improve their understanding of difficult-to-read theoretical articles. Last, an ethics course should spend at least one week on AI ethics, where the fundamentals of how AI models function are reviewed. Students could also practice using that knowledge to critique the benefits and harms of current AI systems and their outputs. The goal for each of the above would be to provide students with foundational awareness of how to use AI tools ethically based on a basic understanding of how AI works. Faculty teaching in a doctoral program could likely follow the suggestions from the preceding paragraph. Further, doctoral programs likely have more robust classes about research methods and quantitative analyses. Because doctoral programs are often aimed at producing leaders within the field, these courses provide an opportunity to dig deeper into the fundamentals of mathematical modeling, AI tools, strategic leadership, governance around AI tools, and how to test and integrate AI tools in behavior analysis. And, for students curious and inclined, doctoral programs provide them the time to build technical fluency in AI to help advance the intersection of AI and behavior analysis conceptually and technically aligned with behavior analysis. Here, students would likely benefit from taking courses from the departments of statistics (e.g., introduction to statistics, probability theory), computer science (e.g., fundamentals of programming, best practices in software development), biology (e.g., computational biology, modeling in ecology), and linguistics (e.g., introduction to natural language processing). As an alternative, doctoral programs may consider adding an “analytics specialization” to their course sequence that covers everything AI-related in a manner specific to what their behavior-analytic students need. 4 Those who have graduated will likely need to seek the necessary knowledge via continuing education (CE). Presentations and workshops around AI have been increasingly offered at conferences or as stand-alone CE events. However, many lack the technical information needed to be AI literate, and it is doubtful a 1-h CE event can provide someone with the skills they need to ethically use AI. Thus, those unaffiliated with university programs likely need to find continuing education coursework or events that teach the necessary skills. Table 3 provides example courses relevant to different AI literacy levels readers might use for CE or faculty might use for content. The courses listed are primarily from Coursera, which offers free access to audit the courses. Of course, this list is not comprehensive and other platforms and alternative courses exist. However, they should provide readers with examples of the types of skills and coursework likely needed for the role(s) they are in or wish to pursue. Table 3. Example Courses for Those Who Have Graduated or Faculty Seeking Examples for Content to Add or Use Level Example Courses Foundational Awareness • Intro to AI (IBM) • Data Privacy Fundamentals (Northeastern University) • AI for Everyone (DeepLearning.AI) Applied Proficiency • Prompt Engineering for ChatGPT (Vanderbilt University) • Introduction to Statistics (many) Analytical & Ethical Integration • Trustworthy AI (many) • Generative AI in the Workplace (University of Michigan) Technical Fluency • Machine Learning Specialization Sequence (Coursera) • CS229 Machine Learning (Stanford Online) • Foundations of Computational and Systems Biology (MIT OpenCourseWare) Strategic Leadership & Governance • AI Strategy and Governance (University of Pennsylvania) • Developing AI Policy (University of London) Open in a new tab In addition to formal courses, it is important to note that many behavior-analytic resources exist around mathematically modeling behavior–environment relations. For example, several excellent tutorials and explanatory guides outline how mathematical modeling of behavior–environment relations works (Critchfield & Reed, 2009 ; Dallery & Soto, 2013 ). A special section in this journal shows the many ways modeling has applied relevance (Jarmolowicz et al., 2022 ). And book-length descriptions of statistical modeling for behavior analysts provide a more thorough step toward understanding some of the fundamental concepts (Cox & Vladescu, 2023 ). AI systems are certainly more complicated, involve many more independent variables, sometimes more dependent variables, and often a heck-of-a-lot more data. However, the same fundamental ideas behind building and evaluating mathematical models from the quantitative analyses of behavior are used to build and evaluate the mathematical models labeled as AI. Summary AI systems will likely be a part of our everyday lives forever. That is inescapable. As with any tool or technology, understanding how to use AI systems ethically requires that the end-user know whether they need the AI tool, which tool best suits their use-case, how to evaluate the outputs of an AI system critically, and the conditions under which an AI system should and should not be used. In an ideal situation, much of this information will come directly from the AI system developers, easing the burden of such analysis on the AI system end-user. But, until regulation requires AI system developers to do so, each behavior analyst will need to become AI literate to conduct this analysis independently, as commensurate with their job roles. Over time, a formal, empirically backed outline of AI literacy and the associated topographies specific to professional behavior analysts will be developed. In the interim, behavior analysts who learn how mathematical model-building works will jumpstart their path toward AI literacy because it fundamentally underpins how AI systems work and AI literacy skills. Funding None. Data Availability There is no data associated with this article. Compliance with Ethical Standards Competing Interests None. Clinical Trial Number None. Footnotes 1 I recognize that this is a utilitarian framing of the question. Which normative ethical framework one prefers is not central to this article. Thus, the reader can likely bend this question toward their preferred ethical theory and the meat and potatoes below remain similarly seasoned. 2 Hereafter, reference to something as ethical or unethical is defined via the utilitarian framing described above as constrained by deontological obligations present in the code of ethics specific to a reader’s profession (e.g., Behavior Analyst Certification Board [BACB], 2020 ). Instead of writing out that many-worded phrase each time, I simply use ethical or unethical to make the writing easier to read. But it should not be read as removing the burden of conducting a complete ethical analysis when the conversation moves from published article to claims about specific behavior in everyday settings. 3 It is important to note is that this is an emerging literature and likely to change as the AI landscape changes over time. 4 At Endicott, we have done this in a three-part sequence via (1) group design class that introduces students to common statistics; (2) introduction to scientific programming in Python; and (3) introduction to mathematical modeling in behavior science. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Baldwin, T. T., & Ford, J. K. (1988). Transfer of training: A review and directions for future research. Personnel Psychology, 41 (1), 63–105. 10.1111/j.1744-6570.1988.tb00632.x [ Google Scholar ] Behavior Analyst Certification Board. (2020). Ethics code for behavior analysts. https://bacb.com/wp-content/ethics-code-for-behavior-analysts/ Bransford, J. D., Brown, A. L., & Cocking, R. R. (2000). How people learn: Brain, mind, experience, and school . National Academy Press. [ Google Scholar ] Celik, I. (2023). Exploring the determinants of artificial intelligence (AI) literacy: Digital divide, computational thinking, cognitive absorption. Telematics & Informatics, 83 (102026), 1–11. 10.1016/j.tele.2023.102026 [ Google Scholar ] Critchfield, T. S., & Reed, D. D. (2009). What are we doing when we translate from quantitative models? The Behavior Analyst, 32 (2), 339–362. 10.1007/BF03392197 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Cox, D. J. (2021). Descriptive and normative ethical behavior appear to be functionally distinct. Journal of Applied Behavior Analysis, 54 (1), 168–191. 10.1002/jaba.761 [ DOI ] [ PubMed ] [ Google Scholar ] Cox, D. J., & Jennings, A. M. (2023). The promises and possibilities of artificial intelligence in the delivery of behavior analytic services. Behavior Analysis in Practice, 17 (1), 123–136. 10.1007/s40617-023-00864-3 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Cox, D. J., & Vladescu, J. C. (2023). Statistics for applied behavior analysis practitioners and researchers . Elsevier. [ Google Scholar ] Dallery, J., spsampsps Soto, P. L. (2013). Quantitative description of environment-behavior relations. In G. J. Madden, W. V. Dube, T. D. Hackenberg, G. P. Hanley, spsampsps K. A. Lattal (Eds.), APA handbook of behavior analysis, vol. 1: Methods and principles (pp. 219–249). American Psychological Association. 10.1037/13937-010 Domínguez Figaredo, D., & Stoyanovich, J. (2023). Responsible AI literacy: A stakeholder-first approach. Big Data & Society, 1–15. 10.1177/20539517231219958 Fauzihardani, E., Rasyid, E. R., Yonedi, E., & Handra, H. (2019). Ethical knowledge, ethical decision: A gap of ethical behavior development. In Proceedings of the 2nd Padang International Conference on Education, Economics, Business and Accounting (PICEEBA-2 2018). 10.2991/piceeba2-18.2019.2 Goldman Sachs. (2023). AI investment forecast to approach $200 billion globally by 2025. https://www.goldmansachs.com/insights/articles/ai-investment-forecast-to-approach-200-billion-globally-by-2025 Haenlein, M., & Kaplan, A. (2019). A brief history of artificial intelligence: On the past, present, and future of artificial intelligence. California Management Review, 61 (4), 5–14. 10.1177/0008125619864925 [ Google Scholar ] Jarmolowicz, D. P., Greer, B. D., Killeen, P. R., & Huskinson, S. L. (2022). Applied quantitative analysis of behavior: What it is, and why we care—introduction to the special section. Perspectives on Behavior Science, 44 (4), 503–516. 10.1007/s40614-021-00323-w [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Jennings, A. M., & Cox, D. J. (2024). Starting the conversation around the ethical use of artificial intelligence in applied behavior analysis. Behavior Analysis in Practice, 17 (1), 107–122. 10.1007/s40617-023-00868-z [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Johnson, E. J., & Payne, J. W. (1985). Effort and accuracy in choice. Management Science, 31 (4), 395–414. 10.1287/mnsc.31.4.395 [ Google Scholar ] Kish-Gephart, J. J., Harrison, D. A., & Treviño, L. K. (2010). Bad apples, bad cases, and bad barrels: Meta-analytic evidence about sources of unethical decisions at work. Journal of Applied Psychology, 95 (1), 1–31. 10.1037/a0017103 [ DOI ] [ PubMed ] [ Google Scholar ] Lintner, T. (2024). A systematic review of AI literacy scales. NPJ Science of Learning, 9 (1), 50. 10.1038/s41539-024-00264-4 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Ng, D. T. K., Leung, J. K. L., Chu, S. K. W., & Qiao, M. S. (2021). Conceptualizing AI literacy: An exploratory review. Computers & Education: Artificial Intelligence, 2, Article 100041. 10.1016/j.caeai.2021.100041 O’Neil, C. (2016). Weapons of math destruction: How big data increases inequality and threatens democracy . Crown. Payne, J. W., Bettman, J. R., & Johnson, E. J. (1988). Adaptive strategy selection in decision making. Journal of Experimental Psychology: Learning, Memory, & Cognition, 14 (3), 534–552. 10.1037/0278-7393.14.3.534 [ Google Scholar ] Reynolds, S. J., Leavitt, K., & DeCelles, K. A. (2010). Automatic ethics: The effects of implicit assumptions and contextual cues on moral behavior. Journal of Applied Psychology, 95 (4), 752–760. 10.1037/a0019411 [ DOI ] [ PubMed ] [ Google Scholar ] Tan, Q., & Tang, X. (2025). Unveiling AI literacy in K-12 education: A systematic literature review of empirical research. Interactive Learning Environments , 1–17. 10.1080/10494820.2025.2482586 Turgeon, S., & Lanovaz, M. J. (2020). Tutorial: Applying machine learning in behavioral research. Perspectives on Behavior Science, 43 , 697–723. 10.1007/s40614-020-00270-y [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Walter, Y. (2024). Embracing the future of artificial intelligence in the classroom: The relevance of AI literacy, prompt engineering, and critical thinking in modern education. International Journal of Educational Technology in Higher Education, 21 , 15. 10.1186/s41239-024-00448-3 [ Google Scholar ] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Data Availability Statement There is no data associated with this article. Articles from Perspectives on Behavior Science are provided here courtesy of Association for Behavior Analysis International ACTIONS View on publisher site PDF (633.2 KB) Cite Collections Permalink RESOURCES Similar articles Cited by other articles Links to NCBI Databases Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top

Related documents

• Literature and Ethical Values#67204
• Ethical issues in multi-agent AI systems for healthcare: a narrative review.#182465
• A sociotechnological-system approach to AI ethics#96972
• The 3D model of ethical AI practice#699971
• From what to how: an initial review of publicly available AI ethics tools, methods and research to translate principles into practices#754949